📖 Documentation

What is DoAnything?

A hosted AI browser agent. You describe a web task in plain language; the AI opens a real browser in the cloud, figures out the steps, and runs them while you watch.

Getting started

1. Create a free account (5 runs/month).
2. From the dashboard, paste a prompt into Quick run.
3. Click Run it — watch the live screenshot view.
4. When the run completes, the result appears below.

Writing good prompts

• Be specific about the goal. "Get the price of X" beats "Check the website".
• State the output format. "Return as a JSON array" or "Return as a numbered list."
• Include the URL if you know it — saves the AI a search step.
• Set max steps sensibly. Most tasks finish in 5–15 steps; raise to 30+ for multi-page work.

Examples that work well

Visit https://news.ycombinator.com — return the title, points,
and URL of the top 5 stories as a markdown table.

Go to https://example.com/pricing — extract each plan name,
monthly price, and the included features. Return as JSON.

Search Google for "best espresso machine 2026" — open the first
result. Summarize the top 3 picks with one sentence each.

What won't work yet (v1.0)

• Tasks requiring login — credential vault arrives in v1.3
• Scheduled / recurring runs — arrives in v1.2
• Webhook delivery — arrives in v1.4 (HookHub integration)
• CAPTCHAs that need solving — task will fail with a clear error

Plans & quotas

See pricing. Quotas reset on the 1st of each month.

Support

Reach the team via sevinhub.com support tickets.

---

Version 1.1.0 · Part of the SevinHub suite